Blog - Investing Notes

July 15, 2008 - Martin Zweig's Four Percent Model Indicator

A reader inquired about the validity of Martin Zweig's Four Percent Model Indicator, which states:

"The Four Percent Model Indicator uses the...weekly close of the Value Line Index. A buy signal is generated when the index rises four percent or more from the previous week. Similarly, a sell signal is indicated when the index falls four percent or more from the previous week."

Does the rule really work? Using weekly closes of the Value Line Arithmetic Index over the period 5/4/84 (the earliest available) through 7/11/08, we find that:

In testing this indicator, we make the following assumptions:

• Buy (sell) the index at the Friday close for any week the index rises (falls) by at least 4%. This timing presumes that we can call the 4% threshold just before the close.
• After a buy (sell) signal, ignore as non-actionable any subsequent buy (sell) signals that occur before the next sell (buy) signal.
• Begin the analysis with the first signal generated during the sample period (a buy on 8/4/84).
• End the analysis on 7/11/08 (while the indicator is on a sell signal).
• Cash earns interest at the 3-month Treasury bill (T-bill) yield while out of stocks.
• One-way trading friction is roughly 0.2% when entering or exiting stocks.
• Ignore the tax consequences of trading.

The following chart summarizes the returns for the Value Line Arithmetic Index while in and out of stocks for the 17 actionable buy and 17 actionable sell signals over the entire sample period. (There are 21 intervening non-actionable buy signals and 20 intervening non-actionable sell signals.) Trade durations vary considerably. Average time in the market is 62% (779 out of 1248 weeks).

Before trading costs, the average weekly return on stocks while in equities is 0.30%, compared to 0.12% while out of equities. This result indicates that the signal does offer information about future returns. The standard deviation of weekly returns is 1.93% while in equities and 2.41% while out of equities.The gap between average daily returns while in and out of stocks is larger from the beginning of the sample through 1997 (0.36% - 0.15% = 0.21%) than since 1997 (0.23% - 0.08% = 0.15%).

What are average weekly returns for an indicator-based trading history?

Constructing a trading history in accordance with the above assumptions (return on cash equal to the T-bill yield and 0.2% one-way trading frictions) results in an average weekly return of 0.22% with standard deviation 1.54%. By comparison, buying and holding the Value Line Arithmetic Index over the same period offers an average weekly return of 0.23% with standard deviation 2.13%. Using the indicator therefore improves risk-adjusted, but not raw, returns.

How do these results play out for cumulative returns?

The next chart compares the cumulative values of $10,000 initial investments in the Value Line Arithmetic Index for a buy-and-hold strategy and a strategy based on Martin Zweig's Four Percent Model Indicator over the entire sample period. Terminal values are close, with buy-and-hold edging the indicator by about 6%. If we set trading friction to zero, terminal values are nearly identical. Cumulative value calculations are sensitive to the start date.

In summary, a trading strategy based on Martin Zweig's Four Percent Model Indicator beats a buy-and-hold strategy on a risk-adjusted, but not raw, return basis over the past 25 years (ignoring tax implications of trading).

For related research, see Blog Synthesis: Some Trading Indicators.